I think solaris (by default) doesn't allocate nearly enough inodes as
(for example) ext3 inder linux. I recall dealing with inode issues
under solaris 6 and 7 for an email server. I agree- a royal pain......
On my system, a redhat box monitoring 122 servers and with nearly
150,000 history files, looks like this:
Filesystem Inodes IUsed IFree IUse% Mounted on
/dev/sda5 2562240 22092 2540148 1% /
/dev/sda1 127744 86 127658 1% /boot
none 129323 1 129322 1% /dev/shm
/dev/sdb1 3842720 159572 3683148 5% /home
/dev/sda6 895840 48 895792 1% /tmp
/dev/sda3 2562240 151660 2410580 6% /usr
/dev/sda2 2562240 23386 2538854 1% /var
Kent C. Brodie - user-da7f7d5174c0@xymon.invalid
Department of Physiology
Medical College of Wisconsin
(XXX) XXX-XXXX
-----Original Message-----
From: Henrik Stoerner [mailto:user-ce4a2c883f75@xymon.invalid]
Sent: Friday, January 05, 2007 10:42 AM
To: user-ae9b8668bcde@xymon.invalid
Subject: Re: [hobbit] out of inodes
On Fri, Jan 05, 2007 at 01:26:52PM +0100, Longina Przybyszewska wrote:
Hobbit (4.2 Solaris9) made the whole disk partition out of game
because of
"out of inodes" fail.
Disk partition is 3.8Gb , used 1.7 Gb, available 2.1Gb.
The directory /opt/hobbit/data/histlogs/ keeps huge amount of
historical data about each hobbit client.
In my configuration I keep to the default values for history events.
What shall I look after, and which variables should be tuned to avoid
that fail in the future?
Hobbit never deletes history-logs by itself. You can use the
"trimhistory" utility to remove old logfiles, or you can just setup a
simple "find /home/hobbit/data/histlogs -mtime +90 | xargs rm" cronjob
to wipe all history-logs older than 90 days (or whatever you prefer).
Regards,
Henrik